打开外部代码时Rpy2解析错误

Question

I'm having problems in running external R script from python using rpy2 library. The setting is the same of this question but I don't feel it solved my issue. By summarizing the problem, I'm able to run an R function by hard typing it but not but calling the script that contains it since I get a parsing error.

More specifically, I have an R script called 'produce_CI.r' which contains the following code

library(readr)
library(bnlearn)

perform_CI <- function(c1, c2, s, dataset, test_type){
  print(c(c1,c2,s))
  if (!s[1]=="()"){
    t <- ci.test(c1, c2, c(s), data = dataset, test = test_type)
  }else{
    t <- ci.test(c1, c2, data = dataset, test = test_type, debug = TRUE)
  }
  print(t)
  return <- t
}

and I try to call it from python using this code

import rpy2.robjects as robjects
import os

robjects.r("""source(%s)""" % os.path.join(os.getcwd(), 'produce_CI.r'))

which gives me the following parsing error

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "path/ci_venv/lib/python3.8/site-packages/rpy2/robjects/__init__.py", line 437, in __call__
    p = rinterface.parse(string)
  File "path/ci_venv/lib/python3.8/site-packages/rpy2/rinterface_lib/conversion.py", line 45, in _
    cdata = function(*args, **kwargs)
  File "path/ci_venv/lib/python3.8/site-packages/rpy2/rinterface.py", line 103, in parse
    res = _rinterface._parse(robj.__sexp__._cdata, num, rmemory)
  File "path/ci_venv/lib/python3.8/site-packages/rpy2/rinterface_lib/_rinterface_capi.py", line 652, in _parse
    raise RParsingError('Parsing status not OK',
rpy2.rinterface_lib._rinterface_capi.RParsingError: Parsing status not OK - PARSING_STATUS.PARSE_ERROR

The only way I'm able to make the code run is by hard-coding it using a string in python

from rpy2.robjects.packages import SignatureTranslatedAnonymousPackage as STAP

r_code = '''library(readr)
library(bnlearn)

perform_CI <- function(c1, c2, s, dataset, test_type){
  print(c(c1,c2,s))
  if (!s[1]=="()"){
    t <- ci.test(c1, c2, c(s), data = dataset, test = test_type)
  }else{
    t <- ci.test(c1, c2, data = dataset, test = test_type, debug = TRUE)
  }
  print(t)
  return <- t
}
'''

testy = STAP(r_code, "testy")
# ====
# removed code to initialize t,x,z_vec, df_r and r_method
# ====
testy.perform_CI(t,x,z_vec, df_r, r_method)

I'm not satisfied by this solution since I'd like to call directly the code in the R script since then I don't have duplicate code. I'm therefore searching for a solution to this parsing error, not a way to just "make the code work".

I feel like an issue may be that I'm writing the code on MacOS and running it on an Ubuntu machine therefore there may be problems with line endings. Therefore I tried to read the R code with python as it was a string and (obviously) I see lot a '\n' everytime there's a new line.

I'm running this code with python 3.8 and rpy2 version 3.4.5

Answer 1

The line

robjects.r("""source(%s)""" % os.path.join(os.getcwd(), 'produce_CI.r'))

will produce a string like source(/path/to/my/produce_CI.r) to be evaluated as R code. Quotes are missing.